Interactive playgrounds and demos teaching multi-agent system design — with real code, common mistakes, and pattern comparisons from the book.
Source: "Agentic Architectural Patterns for Building Multi-Agent Systems" by Dr. Ali Arsanjani & Juan Pablo BustosLearn how MCP (vertical: agent-to-tools) and A2A (horizontal: agent-to-agent) work together. Explore agent cards, agent servers, SSE streaming, and discovery. Includes OpenClaw as a real-world example.
Build agent hierarchies from Tools to AgentTools to Sub-agents to Orchestrators. See how agents are defined, instructed, and spawned. Drag-and-drop composition with execution simulation and callback tracing.
Step through retrieval-augmented generation one stage at a time. Compare No RAG vs Basic RAG vs Agentic RAG using the book's ProWidget E404 example. See embeddings, retrieved chunks, and assembled prompts.
Run the same task through both architectures simultaneously. Inject failures to see SPOF vs resilience. Toggle checkpointing and structured output. Includes the book's actual LoanOrchestrator and Swarm code.
See how NSAF's 5-stage autonomous app factory maps to the book's patterns: Tool Use, Human-in-the-Loop, Supervisor Architecture, Instruction Fidelity Auditing, and Production Readiness. Inject failures and see why the PoC Trap is dangerous.
Multi-channel A2A gateway with WebSocket routing, message flow animation, and tool approval workflow.
Shared task board with self-selecting agents, handoff chain visualization, and speed controls.
Chat interface comparing RAG-enhanced vs naive responses with retrieval visualization.
Loan approval supervision tree with fault injection, recovery animation, and conflict resolution.
NSAF-style autonomous app factory with idea generation, build queue, and deployment dashboard.